The effect of differences in trial design on estimates of efficacy of olanzapine in randomized studies.

Background: Differences in trial design may affect estimates of efficacy of psychotropic drugs. The purpose of this meta-analysis is to evaluate whether the use of Olanzapine (OLZ) as either investigational or control drug affects the observed efficacy of OLZ.

Methods: We performed a search for Randomized-Controlled Trials (RCTs) in which the efficacy of OLZ is assessed in patients with schizophrenia or schizoaffective disorder. We assessed overall efficacy of OLZ and performed subgroup analyses of studies with OLZ as intervention or comparator. Mixed-effect meta-regression analyses were performed.

Results: Of the 25 RCTs included, OLZ was considered as investigational drug or active control in 13 and 12 studies, respectively. The reduction of PANSS score was greater in trials in which OLZ was used as investigational drug. Multivariate meta-regression models showed that a higher PANSS score at baseline and trial duration were the main predictors of greater PANSS score reduction.

Conclusions: Trials with OLZ used as investigational drug differ from those of trials with OLZ as comparator for baseline PANSS scores and study duration; these differences may produce differences in estimates of efficacy. As a consequence, the severity of illness at enrollment and trial duration should be carefully considered to ensure the reliability of indirect comparisons among antipsychotics.